INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A NUMBER OF EXPERIMENTAL STUDIES on normal subjects (3, 5) and in various groups of patients (1, 2, 4) use the analysis of expired aerosol concentration which is recovered after inhalation of a predefined aerosol bolus. The inspired bolus is a small aerosol volume that ranges between 20 and 300 ml (3, 5). Lung behavior is then usually characterized in terms of aerosol bolus dispersion and/or deposition. Usually, deposition is determined for aerosol boluses delivered at various lung depths by varying the air volume after the aerosol bolus, while keeping preinspiratory lung volume and total inspired volume constant. Deposition is then plotted as a function of penetration volume (Vp), defined as the volumetric difference between the occurrence of inspired aerosol peak and the end of inspiration. For any inhaled bolus, some deposition is likely to occur; therefore, measured deposition should always be a positive number. Nevertheless, reports of experimental deposition in normal subjects sometimes do show negative depositions, e.g., an average of 5% in a group of 12 nonsmoking healthy women for a Vp of 250 ml (5). This contrasts with the average value found by Brand et al. (3) in a group of 79 asymptomatic nonsmoker subjects, i.e., 10% deposition for the same Vp, with no difference between female and male subjects. Although the much smaller depositions (5) could partly be accounted for by the 2.5-fold larger flows used in the study, these still do not explain negative deposition values.

Existing reports of deposition data on normal subjects with the use of comparable aerosol bolus maneuvers and flows show substantial differences. For example, Brand et al. (3) and Anderson et al. (1, 2) report average depositions in healthy subjects of 30, 25, and 45%, respectively, at a Vp of 600 ml. There are only small methodological differences among these groups in terms of aerosol size (0.8 or 1 µm), end-inspiratory lung volume, inspired aerosol bolus volume (20-70 ml), the presence or absence of a dental compound to fill the mouth cavity, and the computation method employed. The large range of normal deposition values measured in these studies suggests another source of variability.

In this study, we show how the aerosol delivery system itself can affect deposition results to the extent that in a normal subject 1) deposition becomes negative at low Vp and 2) deposition is underestimated over a wider Vp range. We tested the hypothesis that deposition is crucially dependent on the shape of the inspired bolus through the way it is delivered and measured. First, inhaled bolus shape was investigated by using a 2-liter syringe to pull a 60-ml aerosol bolus through a photometer, with the aerosol delivery system in different configurations. We then used two configurations that gave the most widely different inspired bolus shapes to show that deposition differences of up to 25% can be obtained in the same normal subject, solely depending on the aerosol delivery system. Finally, we evaluated to what extent the deposition results obtained in the two configurations are affected by the computation method used. On the basis of our observations, we suggest specific aspects of inspired aerosol bolus shape that should be checked in any existing aerosol bolus delivery system.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experimental Configuration

View larger version (29K): [in this window] [in a new window]   Fig. 1.   Schematic representation of aerosol bolus-delivery system in unobstructed configuration, which was used either with both photometer (P1) and particle-counter (P2) measurement or with only photometer P1 measurement. In the cases in which particle counter P2 was used, its sampling tube was positioned either in the center or on the edge of the breathing tube. Alternative system configurations used in this study consisted of introducing different degrees of disturbance to flow between valves 1 and 2 (see System Configurations for details). NR, nonrebreathing valve; EX, expiratory.

The system was designed for volume-controlled aerosol bolus inhalation in human subjects. A typical subject test sequence started with some tidal breaths, i.e., clean air inhalation from the box (via NR valve and valve 1) and exhalation to the expiratory bag (via valve 1 and NR valve). Simultaneously, the aerosol was generated into an open circuit through the aerosol tube between valves 3 and 2, with a volume of 60 ml between them. The actual bolus test was then started with clean air inhalation from functional residual capacity (FRC) until a predefined volume was reached, and valves 1, 2, and 3 were reversed, so that the aerosol volume between valves 2 and 3 is inhaled. This was followed by clean air beyond valve 3 from the box until the integrated pneumotachograph signal reached the preset inspired volume. The subject then exhaled to residual volume into the exhalation bag. In the human bolus tests, the preset inspired volume was 1 liter at all times, and the inspiratory pathway was not blocked at end inspiration.

To minimize inspiratory particle losses, valves and connecting tubes were designed such that the lumen of the aerosol pathway (valve 3 to photometer P1) was large and of constant diameter throughout (22 mm ID). The volume between valves 2 and 1 was 15 ml. The volume between valve 1 and photometer P1 was 25 ml, with the sampling tube for the particle counter P2 between them. The mouthpiece was directly connected to the photometer P1.

Aerosol-Measurement Devices

The particle counter P2 is a customized version of the commercially available optical particle counter PCS-2000 (PALAS, Karlsruhe, Germany), which uses a white-light source to illuminate a measurement volume through which particles have to move singly. The particle counter P2 samples part of the respired gas stream via a 6-mm-ID sampling tube that can be fitted to sample (at 2 l/min) either from the center or from the edge of the main breathing tube. Inside the particle counter, the aerosol is diluted (10:1) before it passes through the optically defined measurement volumes to count particles up to number concentrations of 10⁶ particles/cm³. The first factory-installed modification incorporated in our particle counter P2 was the addition of an extra measurement volume and an extra photomultiplier for the optimization of measurement in the concentration range below 20,000 particles/cm³. In the present study, aerosol concentration was in this low concentration range, as is usually the case for this type of experiment. The second factory-installed modification was the particle-counter software, which integrated signals from the low- and high-concentration range and ensured continuity between them.

System Configurations

Unobstructed configuration. This corresponds to the configuration depicted in Fig. 1, which provided a straight, smooth-walled path, with 22-mm ID, from valve 3 to the photometer P1.

Coarse-mesh configuration. In this case, a nylon mesh with 1-mm openings was fixed to a 22-mm OD ring which fitted in the connecting tube between valves 1 and 2, immediately adjacent to valve 2. As a consequence, the bolus, when inhaled, immediately passed through the mesh.

Fine-mesh configuration. This is similar to the coarse-mesh configuration, except that the openings in the mesh were 0.04 mm. The mesh originated from a model 3830 pneumotachograph (Hans-Rudolph).

Turbine configuration. A fixed turbine device, taken from a Microloop spirometer (Micro Medical, Kent, UK), was inserted in place of the mesh.

A 90° angle configuration. An elbow tube (ID = 19 mm) was inserted between valves 1 and 2 so that the inhaled bolus was forced to negotiate a 90° bend before passing the photometer P1. In this configuration, there were no other disturbances in the flow path itself.

Experimental Protocol

The study was conducted in two parts.

Syringe tests. Syringe tests were performed by using a 2-liter syringe connected to the photometer, via a filter, to simulate a 2-liter inhalation with release of the aerosol after 1 liter had been inhaled. The operator pulling the syringe was aided, by a visual feedback of flow on the computer screen, to achieve a constant flow rate of 300 ml/s. A single sequence involved three syringe tests, in each of the five system configurations, for a total of 15 tests. One test sequence was performed for each particle size (0.5, 1, and 2 µm). For these syringe-test sequences, only the photometer P1 was used. After intermediate computation of the results, the 1-µm aerosol tests were repeated in the unobstructed and in the 90° angle configurations by using both the photometer P1 and the particle counter P2, with the particle-counter-sampling tube positioned either in the center or on the edge of the breathing tube.

Human tests. A normal subject performed two sequences of 15 aerosol bolus experiments: one sequence in the unobstructed configuration and one in the 90° angle configuration. Each of the subject test sequences consisted of aerosol boluses inhaled between FRC and FRC + 1 liter and was targeted to cover Vp ranging between 150 and 650 ml, alternately performed in increasing or decreasing Vp order. At end-inspiratory lung volume of 1 liter above FRC, the subject immediately exhaled to residual volume. The subject was aided by flow on the computer screen to target inspiratory and expiratory flows at ~300 ml/s. Because the inspiratory pathway was not blocked at 1 liter above FRC, the intended Vp could increase slightly because of further subject inhalation. After each test, it was checked that this 1-liter inspired volume overshoot never exceeded 50 ml, that the subject flows were within a 10% margin around 300 ml/s, and that end-inspiratory breath hold was <1 s. For these experiments, we used the 1-µm-diameter aerosol, monitoring the experiments only with the photometer P1. The subject performed some extra tests in the unobstructed and 90° angle configuration, with a Vp targeted to ~400 ml, this time using both the photometer P1 and the particle counter P2, with the sampling tube positioned either on the edge or in the center of the breathing tube.

Data Analysis

In the case of the syringe tests, the photometer concentration traces were used to compute inhaled-bolus half-width (H_in) by measuring the volumetric width of the photometer signal at one-half peak height. We also determined three bolus volumes representing the volumes which contained 85, 90, and 95% of the total quantity of the bolus (Vol_85%, Vol_90%, and Vol_95%, respectively). For instance, Vol_85% was determined such that

(1)

where C is the aerosol concentration, V is volume, and Vol₀ is the volume corresponding to the onset of the incoming bolus front (i.e., first photometer concentration above background noise level). A large value of these bolus volumes compared with the half-width is indicative of a long tail on the bolus. Vol_90% is directly comparable to the measurement in McNamee and Boykin (7).

In the case of subject aerosol tests, we computed dispersion (H) by using

(2)

where H_in and H_ex are the half-width of inspired and expired bolus, respectively.

Deposition (De) was calculated by using

(3)

Deposition was computed using three different methods: 1) by integration over the entire inspired- and expired-volume range, 2) by integration over the volume range in which C was at least 5% of the peak expiratory aerosol concentration, and 3) by integration over the inspired-volume range in which C was at least 5% of the peak inspiratory aerosol concentration, and over the expired-volume range in which C was at least 5% of the peak expiratory aerosol concentration. The choice of 5% was based on previously reported values ranging from 1 to 15%, depending on the applied method (1-3, 5), and the intentional use of the same percentage for methods 2 and 3.

Statistical Analysis

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Syringe Results

View larger version (10K): [in this window] [in a new window]   View larger version (14K): [in this window] [in a new window]   Fig. 2.   A: inhaled half-widths (Hin; means ± SD) obtained from syringe-bolus tests with aerosol setup in unobstructed (unobstr), coarse-mesh, fine-mesh, turbine, or 90° angle configurations. B: bolus volumes containing 85, 90, and 95% of the bolus (Vol85%, Vol90%, and Vol95%, respectively) obtained from same syringe-bolus tests as in A.

We checked for H_in, Vol_85%, Vol_90%, or Vol_95% differences among the different configurations in Fig. 2. Essentially, H_in was different between any two configurations, except between coarse mesh and fine mesh. In contrast, Vol_85%, Vol_90%, and Vol_95% differed only between unobstructed and coarse mesh, which in turn differed from the remaining three configurations. There was no difference in Vol_85%, Vol_90%, or Vol_95% among fine mesh, turbine, and 90° angle configurations. Note how, in the worst-case scenario presented here (unobstructed), 95% of the aerosol that was originally confined to a 60-ml tube became spread over 450 ml, with a corresponding half-width of only 42 ml. Except for the unobstructed configuration, all other configurations produced inhaled aerosol boluses with 85% of the aerosol contained in ~100 ml (Vol_85% in Fig. 2B). In fine mesh, turbine, and 90° angle configurations, it took another 100 ml to contain an additional 10% of the aerosol (difference between Vol_85% and Vol_95%). The bolus tail spread is evidenced by the large increment between Vol_95% and Vol_90% vs. that between Vol_90% and Vol_85%.

For the two most widely different configurations in terms of H_in and Vol_85%, Vol_90%, and Vol_95% in Fig. 2 (i.e., the unobstructed and 90° angle configurations), both particle counter P2 and photometer P1 measurements are shown in Fig. 3 (dotted and solid lines, respectively) . Figure 3 shows the result of positioning the particle counter P2 sampling tube either in the center (A and C) or on the edge (B and D) of the breathing tube. The mismatch between photometer P1 and particle counter P2 traces in the unobstructed configuration (Fig. 3, A and B) contrasts with the match between corresponding signals in the 90° angle configuration (Fig. 3, C and D). In the unobstructed configuration, the degree of mismatch differs, depending on whether the particle counter is sampling from the center or from the edge of the breathing tube. In the 90° angle configuration, photometer P1 and particle-counter P2 traces coincide, regardless of whether the particle counter is sampling from the center or the edge of the breathing tube.

View larger version (23K): [in this window] [in a new window]   Fig. 3.   Comparison of photometer (P1; solid line) and particle-counter (P2; dotted line) measurements of syringe-bolus test with setup in unobstructed (A and B) and 90°angle (C and D) configurations and with P2-sampling tube positioned in center (A and C) or at edge (B and D) of breathing tube.

Subject Results

View larger version (18K): [in this window] [in a new window]   View larger version (15K): [in this window] [in a new window]   Fig. 4.   A: bolus-dispersion values (H; means ± SD) obtained in same subject with setup in unobstructed (open bars) and 90° angle (solid bars) configurations. Vp, penetrating volume. B: bolus-deposition values (De; means ± SD) obtained in same subject with setup in unobstructed (open bars) and 90° angle (solid bars) configurations. Deposition values shown in this figure are computed by using method 1, i.e., performing integrations in Eq. 3 over entire inspiration and entire expiration. * Significant differences between unobstructed and 90° angle configuration, P < 0.05.

Table 1 (top) shows the mean values (±SD) of all deposition results obtained by using the three computation methods (methods 1, 2, and 3). In particular, the values listed under deposition, method 1, in Table 1 are those corresponding to the data points in Fig. 4B. Table 1 (bottom) also shows the average difference between deposition values obtained in 90° angle and unobstructed configurations as a function of Vp. Table 1 (top) shows the greatest difference between depositions computed with methods 2 and 3. Table 1 (bottom) demonstrates how all three computation methods show the same general pattern, i.e., a smaller dependence of configuration on deposition differences for the larger Vp levels.

Finally, inhaled and exhaled aerosol behavior in the unobstructed and 90° angle configurations through the comparison of photometer P1 and particle counter P2 traces obtained on the subject for an aerosol bolus test with an intermediate Vp of ~400 ml can be summarized as follows. Aerosol boluses inhaled by the subject in the unobstructed configuration (Fig. 5) presented the same P1-to-P2 mismatch as the syringe bolus in the unobstructed configuration (Fig. 3A). Aerosol boluses inhaled by the subject in the 90° angle configuration (not shown) reproduced the agreement between the photometer P1 and the particle counter P2 that was obtained with the syringe (Fig. 3, C and D). However, in both configurations, there was a good agreement between expiratory P1 and P2 concentration traces regardless of whether the P2 sampling tube was fitted to the center (Fig. 5) or the edge of the breathing tube.

View larger version (10K): [in this window] [in a new window]   Fig. 5.   Example of aerosol bolus inhalation and exhalation in a subject, plotted as a function of cumulatively inhaled and exhaled volume, with both photometer P1 and particle-counter P2 measurements. This test corresponds to a 1-liter inhalation from functional residual capacity, with Vp of 400 ml, followed by an exhalation down to residual volume, which in this case was 1.5 liters below functional residual capacity.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this study, we illustrate how measurement errors of the inspired aerosol bolus can lead to severe experimental underestimation of aerosol bolus deposition. This type of experimental underestimation is present at all Vp values and is the greatest for low Vp (Fig. 4B), an observation which holds for all three deposition-computation methods used here (Table 1). Although deposition measurement appears to be highly sensitive to the method of inhaled aerosol bolus delivery, this is not the case for aerosol bolus dispersion in terms of H, which was hardly affected by inhaled bolus shape (Fig. 4A).

To provide an insight into the causes which contribute to the observed differences between the two most widely differing configurations, we compared simultaneous recordings of the syringe boluses with an in-line photometer P1 and on-line sampling particle counter P2 (Fig. 3). Figure 3, C and D, shows that, in the 90° angle configuration, the time course of the aerosol concentration measured by the photometer P1 matched the time course of the aerosol concentration measured by particle counter P2, regardless of the radial position of the P2-sampling tube in the main breathing tube. In contrast, in the unobstructed configuration (Fig. 3, A and B), the mismatch between the photometer P1 and the particle counter P2 suggests a parabolic flow profile of the incoming bolus. This is plausible, because we are dealing with Reynold's numbers of ~1,000 (flow, ~300 ml/s; tube diameter, 2.2 cm). Figure 3 clearly indicates that, under these flow conditions, the aerosol bolus pathway from reservoir to photometer should contain some degree of disturbance to mix the aerosol with the surrounding air so that the aerosol front becomes blunt enough and the photometer P1 measurement may indeed be considered representative of aerosol concentration over the entire tube cross section.

Although the comparisons of photometer P1 and particle counter P2 traces in Fig. 3 could actually provide an explanation for the inadequacy of the photometer measurement, depending on the configuration of the aerosol delivery system, detailed analysis of the photometer traces themselves also reveals distinct differences. For instance, the 90° angle configuration provided a combination of relatively high H_in (Fig. 2A) and relatively low volumetric confinement of the aerosol in terms of Vol_85%, Vol_90%, and Vol_95% (Fig. 2B) in contrast to the unobstructed configuration. These results suggested that the photometer may underestimate actual inhaled aerosol concentration, depending on the configuration of the aerosol delivery system. To further test this hypothesis, we used a relative bolus measurement by having a normal subject perform bolus inhalations. This provides a measure of the magnitude of the experimental error on bolus parameters with clinical relevance.

The tests in a normal subject revealed that, with inhaled boluses delivered in both unobstructed and 90° angle configuration, the photometer and particle-counter traces were matched during subject exhalation, regardless of whether the particle-counter-sampling tube was in the center or on the edge of the breathing tube. This is shown by Fig. 5, representing the worst case, in which the inhaled bolus reproduced the mismatch between the photometer and the particle counter that was previously seen in the syringe tests. In contrast, the exhaled bolus showed no differences between the photometer and the particle counter traces. This suggests that, during exhalation, the photometer measurement is representative of the entire tube cross section, probably because the glottis and the mouth cavity provide sufficient disturbance to blunt the aerosol-flow profile. If we accept that an inadequately delivered bolus primarily affects the recorded inhaled bolus shape and not the exhaled bolus, it follows that overall bolus dispersion will not be affected. Because the bolus H_ex is usually at least triple that of the bolus H_in over the Vp range under study, the effect of an underestimation of the bolus H_in of a few milliliters will be marginal and will be largely overridden by the variability of dispersion measurement itself. Figure 4A confirms this experimentally.

The subject experiments shown in Fig. 5 also indicated that, when the aerosol-flow profile is insufficiently blunted as the aerosol bolus enters the photometer, this leads to an underestimated inhaled-aerosol quantity, whereas exhaled-aerosol quantity measurement is unaffected. This implies that deposition (De in Eq. 3) will be underestimated. Also, a given absolute underestimation of the inspired bolus quantity is expected to provide different degrees of deposition underestimation, depending on actual deposition in the lungs and, therefore, Vp level (with the greater underestimation occurring at low values of Vp). Figure 4B indeed shows that experimental deposition values are generally smaller in the unobstructed than in the 90° angle configuration and that the absolute deposition difference between both configurations is indeed greater for Vp = 200 ml than for Vp = 600 ml.

Table 1, bottom, shows that the pattern of larger deposition difference between configurations for smaller Vp holds for all three computation methods used. It would be expected that the computation methods which discard part of the bolus tail, such as method 2 or 3, would produce quite different deposition values, because the system configuration has a marked effect on the shape of the inspired-bolus signal in terms of volumetric confinement of the bolus tail (Fig. 2B). Table 1, top, clearly shows that the differences in deposition values, in particular between methods 1 and 3, are substantial in the unobstructed configuration but virtually disappear in the 90° angle configuration. From this result, we may conclude that, provided the aerosol setup delivers proper quality of the inhaled bolus, the deposition-computation methods presented here will produce comparable results.

The comparison of aerosol bolus data obtained in different laboratories by using slightly different experimental and computational methodologies is difficult and should be handled with caution when the reported normal values are used as a reference (3). The genesis of errors in deposition, in which the underestimation sometimes becomes apparent through the negative values for small Vp values, may actually be quite subtle. In previous papers, inhaled-aerosol bolus characteristics are usually restricted to the mention of bolus half-width, with a general tendency to decrease bolus H_in as much as technically possible [e.g., 20 ml in Brand et al. (3)], to target the aerosol bolus to narrower lung volumes at a given lung depth. Although the inhaled bolus may indeed be conveniently quantified by its half-width, Fig. 2 clearly shows how a small H_in can nevertheless be associated with a large volumetric spread of the total amount of aerosol (Vol_85%, Vol_90%, and Vol_95%). Even in the best-case scenario, the bolus may still be quite widely spread, as can be seen from the Vol_90% for the 90° angle configuration, i.e., 120 ml for a 60-ml aerosol reservoir, a result which is similar to that in the study by McNamee and Boykin (7), in which a Vol_90% of 61 ml was obtained for an original aerosol volume of 35 ml.

In a previous work that concerned the use of light-scattering photometry in aerosol medicine, Gebhart et al. (6) stated that "before an aerosol enters the photometer, it has to be thoroughly mixed so that the aerosol concentration passing through the sensitive area of the photometer is representative for the whole cross section of the aerosol channel." To our knowledge, the only precise indication of how to achieve this condition is in a paper by Smaldone et al. (8), in which a figure legend refers to right angles in the setup to "promote mixing in tubing and minimize streaming in the aerosol measurement device." In most other studies, the schematic representation of the aerosol setup suggests that this right-angle strategy is not the only one that could be used. It is, therefore, not unreasonable to think that different kinds of obstructions used in different aerosol laboratories may lead to different qualities of the inhaled bolus and, consequently, to different deposition values.

The question that remains is how to assess whether inhaled aerosol boluses are adequately delivered and measured. We recommend first testing the sensitivity of the inhaled aerosol bolus shape to the insertion of any kind of disturbance or, if possible, a 90° angle between aerosol reservoir and photometer. Syringe tests, such as the ones described here, should be performed by using flows corresponding to those that would be used in experiments on human subjects or in patients. The resulting aerosol boluses, as measured by the photometer, can be evaluated most readily by their H_in. When the entire contents of the aerosol bolus reservoir are delivered, as was the case here, one would like H_in to be as close as possible to the aerosol-reservoir volume. In some cases, the aerosol is not originally contained in a well-defined volume but is sampled from a larger reservoir by a system of rapidly switching valves. In this case, one can simply observe whether H_in increases at all when flow is disturbed in one way or another. In addition to H_in measurement, the quality control of inhaled bolus could also benefit from the evaluation of its volumetric confinement in terms of parameters such as Vol_85% and Vol_95%. In particular, the latter (parameters) provide a means of characterizing the aerosol bolus tail, which remains one of the limitations of bolus delivery experiments in general.